6 perf-report - Read perf.data (created by perf record) and display the profile
11 'perf report' [-i <file> | --input=file]
15 This command displays the performance counter profile information recorded
22 Input file name. (default: perf.data unless stdin is a fifo)
26 Be more verbose. (show symbol address, etc)
30 Only consider symbols in these dsos. CSV that understands
31 file://filename entries.
34 Show the number of samples for each symbol
36 --showcpuutilization::
37 Show sample percentage for different cpu modes.
41 Show per-thread event counters
44 Only consider symbols in these comms. CSV that understands
45 file://filename entries.
48 Only consider these symbols. CSV that understands
49 file://filename entries.
52 Only show symbols that match (partially) with this filter.
56 Only display entries resolved to a symbol.
60 Sort histogram entries by given key(s) - multiple keys can be specified
61 in CSV format. Following sort keys are available:
62 pid, comm, dso, symbol, parent, cpu, srcline, weight, local_weight.
64 Each key has following meaning:
66 - comm: command (name) of the task which can be read via /proc/<pid>/comm
67 - pid: command and tid of the task
68 - dso: name of library or module executed at the time of sample
69 - symbol: name of function executed at the time of sample
70 - parent: name of function matched to the parent regex filter. Unmatched
71 entries are displayed as "[other]".
72 - cpu: cpu number the task ran at the time of sample
73 - srcline: filename and line number executed at the time of sample. The
74 DWARF debuggin info must be provided.
76 By default, comm, dso and symbol keys are used.
77 (i.e. --sort comm,dso,symbol)
79 If --branch-stack option is used, following sort keys are also
81 dso_from, dso_to, symbol_from, symbol_to, mispredict.
83 - dso_from: name of library or module branched from
84 - dso_to: name of library or module branched to
85 - symbol_from: name of function branched from
86 - symbol_to: name of function branched to
87 - mispredict: "N" for predicted branch, "Y" for mispredicted branch
89 And default sort keys are changed to comm, dso_from, symbol_from, dso_to
90 and symbol_to, see '--branch-stack'.
94 A regex filter to identify parent. The parent is a caller of this
95 function and searched through the callchain, thus it requires callchain
96 information recorded. The pattern is in the exteneded regex format and
97 defaults to "\^sys_|^do_page_fault", see '--sort parent'.
101 Only display entries with parent-match.
104 --column-widths=<width[,width...]>::
105 Force each column width to the provided list, for large terminal
110 Use a special separator character and don't pad with spaces, replacing
111 all occurrences of this separator in symbol names (and other output)
112 with a '.' character, that thus it's the only non valid separator.
116 Dump raw trace in ASCII.
118 -g [type,min[,limit],order[,key]]::
120 Display call chains using type, min percent threshold, optional print
123 - flat: single column, linear exposure of call chains.
124 - graph: use a graph tree, displaying absolute overhead rates.
125 - fractal: like graph, but displays relative rates. Each branch of
126 the tree is considered as a new profiled object. +
129 - callee: callee based call graph.
130 - caller: inverted caller based call graph.
133 - function: compare on functions
134 - address: compare on individual code addresses
136 Default: fractal,0.5,callee,function.
140 alias for inverted caller based call graph.
142 --ignore-callees=<regex>::
143 Ignore callees of the function(s) matching the given regex.
144 This has the effect of collecting the callers of each such
145 function into one place in the call-graph tree.
148 Pretty printing style. key: normal, raw
150 --stdio:: Use the stdio interface.
152 --tui:: Use the TUI interface, that is integrated with annotate and allows
153 zooming into DSOs or threads, among other features. Use of --tui
154 requires a tty, if one is not present, as when piping to other
155 commands, the stdio interface is used.
157 --gtk:: Use the GTK2 interface.
168 Load module symbols. WARNING: This should only be used with -k and
173 Don't complain, do it.
175 --symfs=<directory>::
176 Look for files with symbols relative to this directory.
179 --cpu:: Only report samples for the list of CPUs provided. Multiple CPUs can
180 be provided as a comma-separated list with no space: 0,1. Ranges of
181 CPUs are specified with -: 0-2. Default is to report samples on all
185 --disassembler-style=:: Set disassembler style for objdump.
188 Interleave source code with assembly code. Enabled by default,
189 disable with --no-source.
192 Show raw instruction encoding of assembly instructions.
194 --show-total-period:: Show a column with the sum of periods.
198 Display extended information about the perf.data file. This adds
199 information which may be very large and thus may clutter the display.
200 It currently includes: cpu and numa topology of the host system.
204 Use the addresses of sampled taken branches instead of the instruction
205 address to build the histograms. To generate meaningful output, the
206 perf.data file must have been obtained using perf record -b or
207 perf record --branch-filter xxx where xxx is a branch filter option.
208 perf report is able to auto-detect whether a perf.data file contains
209 branch stacks and it will automatically switch to the branch view mode,
210 unless --no-branch-stack is used.
213 Path to objdump binary.
216 Show event group information together.
219 Demangle symbol names to human readable form. It's enabled by default,
220 disable with --no-demangle.
223 Do not show entries which have an overhead under that percent.
228 linkperf:perf-stat[1], linkperf:perf-annotate[1]